README_MODEL_TEMPLATE.md

---

Structure of the directory for storing model files

computer_vision
   |--classification      
       |--{scene_name}       
            |--{scene_name}.om
            |--README.md            
   |--object_detect   
   |--segmentation

Note:

The name of the network model file suffixed with om must be the same as the name of the upper-layer folder.

---

Original Network Link:

//Address of the original network model such as caffe/tensorflow/pytorch ---[mandatory]

Pre-trained Model Link:

//Pre-training network model address such as caffe/tensorflow/pytorch ---[mandatory]

Input Data Description:

//Input image requirements, including the size and format ---[mandatory]

Out Data Description:

//Description of the output data of a model, and the address of the lable file corresponding to the training data set ---[mandatory]

Custom Operator:

• //Indicates whether to include custom operators: Included/Not included ---[mandatory]

• //Custom Operator Name: --- [Optional]

• //URL of the Custom operator code: --- [Optional]

• //Custom Operator Name: --- [Optional]

• //URL of the Custom operator code: --- [Optional]

Note: Add all involved custom operators in the preceding format.

Versions that have been verified:

---[mandatory]

• //Atlas 200
• //Atlas 300